Corporate Social Responsibility as a Buffer in Times of Crisis: Evidence from China’s Stock Market During COVID-19

Abstract

Prior research often portrays Corporate Social Responsibility (CSR) as a coercive institutional force compelling firms to passively conform for legitimacy. More recent studies, however, suggest firms actively pursue CSR to gain sustainable competitive advantages. Yet, how and when CSR buffers firms against adverse shocks of crises remains insufficiently understood. This study addresses this gap by using multiple regression analysis to examine the buffering effects of CSR investments during the COVID-19 crisis, which severely disrupted capital markets and firm valuation. Drawing on signaling theory and CSR literature, we analyze the stock market performance of China’s A-share listed firms using a sample of 2577 observations as of the end of 2019. Results indicate that firms with higher CSR investments experienced significantly greater cumulative abnormal returns during the pandemic. Moreover, the buffering effect is amplified among firms with higher debt burdens, greater financing constraints, and those operating in regions with stronger social trust and more severe COVID-19 impact. These findings are robust across multiple robustness checks. This study highlights the strategic value of CSR as a resilience mechanism during crises and supports a more proactive view of CSR engagement for sustainable development, complementing the traditional legitimacy-focused perspective in existing literature.

1. Introduction

Research on Corporate Social Responsibility (CSR) has gained significant momentum [1,2,3,4]. CSR is defined as “corporate actions oriented towards the welfare of stakeholders and driven by instrumental, relational or ethical concerns (p. 3)” [5]. A major stream of CSR research focuses on its implications for firm performance, which can be largely categorized into two streams. The first stream, grounded in the institutional perspective, adopts a passive view and conceptualizes CSR engagement as reactive institutional compliance [6,7]. Such efforts, though, often come at the expense of economic interests and exert a negative impact on firm performance [8,9,10]. In contrast, another group of researchers draws on the resource-based view, portraying firms as proactive agents that invest in CSR to generate firm-specific advantages. From this standpoint, CSR is a strategic resource that can enhance a firm’s sustainable competitive position, implying a positive relationship between CSR and performance [11,12]. Consequently, the existing literature offers mixed findings, leaving the impact of CSR on firm performance ambiguous and insufficiently understood [13]. While conventional perspectives suggest that CSR and sustainability initiatives primarily arise in stable institutional contexts with abundant resources [14,15], their persistence under adverse environmental conditions presents a compelling paradox [16,17,18]. Recent research has begun to explore this counterintuitive phenomenon [19,20], often drawing on data from developed markets to assess the effects of CSR investments on market performance [21,22], risk management [23,24], disclosure quality [25], and corporate governance [26]. Some studies argue that CSR functions strategically as a buffer against external shocks, highlighting its positive role in times of crisis [27]. Conversely, other empirical findings challenge this view, indicating that CSR and sustainability investments may have a limited impact during sudden disruptions [28,29,30,31]. We argue that these inconsistent results highlight the complex, multifaceted mechanisms through which CSR operates, as well as the critical role of internal and external contingencies in shaping these mechanisms. Consequently, a more pressing and nuanced question concerns how CSR serves as a resilience mechanism in adverse environments and when it proves most effective.

This study leverages the COVID-19 pandemic as a crisis context well-suited for examining CSR investments, given its severity, data availability, and timeliness [32,33,34,35]. Drawing on signaling theory [36], we argue that under conditions of information asymmetry, firms’ CSR investments serve as both visible and credible signals to stakeholders, enhancing the quantity (visibility) and quality (credibility) of information. These signaling effects can function as a buffer during crises. The cushioning effect is expected to be most pronounced when firms face strategic imperatives for credible signaling and contextual pressures for visible signaling, particularly among those with higher debt levels, greater financing constraints, and those operating in regions characterized by stronger social trust and more severe COVID-19 impacts.

This study employs event analysis using cross-sectional data from Chinese A-share listed companies during the COVID-19 period, combining CSR scores sourced from Hexun.com with financial indicators from the CSMAR database. We found our hypotheses all supported, thus underscoring the strategic value of CSR in buffering adverse external events. By doing so, we aim to make the following contributions. Firstly, our study advances the literature by integrating signaling theory with crisis management literature to systematically uncover the proactive strategic significance of CSR in buffering the adverse impacts of crises, transcending the conventional passive view of CSR as merely a legitimacy tool. Secondly, our study elucidates the mechanism through which CSR works during pandemic crises, clarifying its transmission channel in enhancing market confidence through strengthening information transparency, thus answering the research question of “how CSR serves as a resilience mechanism in adverse environments”. Thirdly, it further demonstrates that CSR proves most effective at the intersection of internal strategic imperatives and external environmental pressures. These dual contingencies activate CSR’s signaling power, extending signaling theory’s applicability to contexts where non-financial disclosures operate as critical resilience proxies during systemic crises and answering the research question of “when CSR proves most effective”. Our study also offers practical implications. For firms, this study highlights the strategic value of CSR investments as a proactive resilience mechanism during crises. By leveraging CSR as a credible and visible signal, firms can enhance stakeholder confidence, mitigate adverse market reactions, and strengthen their crisis resilience. For policymakers, the findings underscore the importance of fostering institutional environments that incentivize firms’ CSR engagement, especially during systemic crises. Policies promoting CSR disclosures—such as tax incentives, public recognition, or integration into corporate governance frameworks—can contribute to sustaining market stability and facilitating broader socioeconomic recovery.

The work is structured as follows. First, we present the theoretical background and hypotheses based on signaling theory. Next, we outline methods and results. The final part of the work includes a discussion of the results, conclusion, limitations, and future research.

2. Theoretical Background and Hypotheses Development

Signaling theory, initially proposed by Spence [37] in labor market research, addresses adverse selection caused by information asymmetry. The core argument of signaling theory lies in using costly and observable actions to convey unobservable organizational traits or strategic intentions. For example, High-quality employees sent the costly signal of high education to potential employers to show a sustainable competitive advantage [38].

CSR is an effective way to build trust by conveying a credible and visible signal. In other words, although CSR is not a resource in itself, it is an important way for firms to obtain a variety of resources and a favorable business environment from the outside, thereby increasing their sustainable competitive advantage [39]. Trust fundamentally operates through dual signaling properties of visibility and credibility [40,41]. The establishment of trust is inseparable from social capital. Putnam [42,43] views social capital as a tendency of people in a society to cooperate in order to produce socially efficient outcomes and emphasizes the “norms of reciprocity and trustworthiness” that result from connections between individuals. CSR plays the role of information transfer between enterprises and stakeholders [44,45]. And then enterprises build social networks and acquire social capital through the credible and visible signal of CSR, which promotes trust between firms and stakeholders [46,47]. In this way, enterprises enable stakeholders to obtain information about their business status, financial performance, and moral characteristics [48,49].

The positive impacts of CSR investments as a visible and credible signal are analyzed in this work, starting with two key categories of stakeholders: customers and investors. From the customer’s perspective, the behavior of CSR investments by enterprises can be understood as a differentiation strategy, which conveys visible and credible signal to existing and potential customers through the use of environmentally friendly materials, the implementation of charitable donations, and other actions, and creates a differentiated image of the product, which in turn affects consumers’ perception of the quality of the product and the intention to purchase [50]. CSR investments signal a strengthened corporate reputation by aligning brand actions with societal values [51]. This reputation-driven trust cultivates customer satisfaction and loyalty, which reduces demand elasticity and grants firms pricing power to achieve premium margins. Albuquerque et al. [52] empirically demonstrate that such loyalty-driven revenue stability lowers operating leverage, thereby decreasing systemic risk exposure and increasing firm valuation.

From an investor’s perspective, CSR investments signal function as a commitment to long-term sustainability that attracts capital from investors. Studies have shown that investors with a preference for ESG stocks are more focused on sustainability, have lower sensitivities relative to other investors, and have more resilient investment behavior [53,54]. In the COVID-19 pandemic, many investors will sell their stocks and flee the market out of uncertainty, whereas the segment of investors with a preference for ESG will be more resilient and will not choose to sell easily, thus keeping the price of ESG stocks relatively stable [20].

Overall, CSR investments by firms (e.g., ensuring the welfare of workers, providing safe products, honoring informal agreements with suppliers, and protecting the environment) convey a credible and visible signal to stakeholders, which reveals the fulfillment of formal or informal sustainable commitments. These activities can strengthen ties and reduce information asymmetry between firms and their stakeholders (employees, customers, suppliers, investors, government, society, etc.) and form a social foundation of trust. In turn, these stronger bonds can help retain high-quality employees and help maintain stable supply chains and customers in difficult times, while investors’ trust in the firm itself leads to more stable investment sentiment in the face of unexpected events. In other words, firms are able to build social capital and maintain information transparency through CSR investments, which motivates stakeholders to remain loyal and helps firms overcome the challenges of crises when unexpected events occur. Stakeholders will perceive firms with high social capital as having a lower probability of violating the contract due to shared values and cooperative norms. The more a crisis occurs that leads to a decrease in the overall level of trust in society, the more significant the role of social capital becomes [55], helping enterprises to maintain their own stability under the crisis and safely get through it [56]. Therefore, from the perspective of risk management, the action of enterprises to make CSR investments actually plays a role similar to that of insurance, which can effectively reduce the possibility of enterprises suffering from negative crisis impacts. From this, it can be reasonably inferred that firms with more CSR investments will have more positive market performance in the event of emergencies such as COVID-19. Based on the above analysis, we propose the first hypothesis of this work.

Hypothesis 1.

Firms with higher levels of CSR investment before the onset of the COVID-19 pandemic exhibit stronger cumulative stock market performance during the pandemic period.

We now examine the internal contingencies, specifically strategic imperatives, to address the when question. Prior research has shown that firms facing financing constraints often use signals to reduce information asymmetry and improve access to external capital. Czarnitzki and Hottenrott have indicated that patents serve as quality signals that can help small firms alleviate financing constraints [57]. Managerial skill also acts as a signal. Highly skilled managers are better able to communicate their firm’s prospects, thereby reducing information asymmetry and enabling greater debt financing, especially in environments with tight financial resources and high growth opportunities [58]. During the COVID-19 pandemic, firms with prior credit constraints also resorted to trade credit or government grants, which conveyed a credible signal to stakeholders [59]. Overall, for firms grappling with excessive debts or high financing constraints, the cultivation of strategic imperatives for credible signal formation becomes critical, whether through patents, managerial reputation, or other means. CSR investments outperform in mitigating information asymmetry by conveying a credible signal. These firms operate in a small pond of limited financial flexibility, where the need for credible resource replenishment (e.g., access to capital, stakeholder trust) becomes critical. There have been studies confirming that the firms with more cash and fewer liabilities have a more positive market performance during the period of COVID-19 from the point of view of the financial policies [60]. Firms with excessive debts are conceivably more financially inflexible and face a higher risk of bankruptcy, and thus these firms can reasonably be expected to be more vulnerable to the effects of COVID-19 and face higher risks [61].

This risk factor can be mitigated to a certain extent if the degree of social responsibility commitment of enterprises with excessive debts is high. CSR participation strengthens the trust between the enterprise and its stakeholders, and this trust bond can effectively reduce the cost of capital through the mechanism of reducing the degree of information asymmetry, which can help the enterprise to be free from crisis [62]. CSR expenditures provide insurance-like protection that allows firms to reduce volatility in the event of an outbreak in order to maintain firm value. In contrast, for firms with debt levels below optimal leverage, they have sufficient liquidity for working capital in the event of an epidemic outbreak, and the insurance effect from social responsibility may be relatively insignificant. Based on the above analysis, we propose the second hypothesis of this work.

Hypothesis 2.

The positive relationship between pre-pandemic CSR investments and cumulative stock market returns will be stronger for firms with higher levels of debt or more financing constraints.

We now look for external contingencies from the angle of contextual pressures. For firms located in regions with higher social trust and more severe pandemic impact, contextual demands for visible signal enhancement become critical. Firms in regions with high social trust and severe pandemic impacts face intensified competition-induced uncertainty (e.g., stakeholder skepticism, opaque information environments). These regions represent a big pond of competitive signaling, where CSR’s visibility determines market access. The efficacy of CSR as a visible signal is contingent on two contextual amplifiers: regional social trust and pandemic severity. Firms located in regions with higher social trust and more severe pandemic impacts will be more affected by information opacity in the stock market returns. Under such scenarios, CSR investments act as a visible signal, cutting through the information fog. High social trust environments enhance signal efficiency by reinforcing social norms (e.g., reputation constraints), reducing stakeholder skepticism about CSR motives (e.g., greenwashing risks) [63]. Simultaneously, pandemic severity intensifies information demands, compelling stakeholders to rely on CSR actions to infer firms’ crisis management capabilities. Based on the above analysis, we propose the third hypothesis of this work. The research framework is illustrated in Figure 1.

Hypothesis 3.

The positive relationship between pre-pandemic CSR investments and cumulative stock market returns will be stronger for firms located in regions characterized by higher levels of social trust or more severe COVID-19 impact.

3. Method

3.1. Research Design

To explore the buffering effect of CSR investments on the impact of the COVID-19, this work utilizes the event study method to conduct the analysis, using the cumulative abnormal returns within the event window to reflect the changes in stock returns of listed companies before and after the outbreak of the COVID-19, and to reflect the impact of the COVID-19, an exogenous event, on the research subjects. The first officially notified case of COVID-19 developed on 8 December 2019, followed by the issuance of the Emergency Notice on the Rescue and Treatment of Unexplained Pneumonia by the Medical Affairs and Medical Management Division of the Wuhan Health Commission on 30 December, and the arrival of experts from the National Health Commission in Wuhan to carry out the relevant testing and verification work on 31 December 2019, which was the day when the Wuhan Health Commission first notified of the outbreak of pneumonia, with no apparent phenomenon of human-to-human transmission [64]. On 11 January 2020, there were 41 cases of infection; at this time, Wuhan’s notification still pointed out that no medical personnel were found to be infected, and it is not found that there is obvious evidence of human-to-human transmission [65]. Up to this point, although there have been continuous reports of the pneumonia epidemic, at that time, its extremely contagious hazards had not yet been exposed, and did not cause people to pay enough attention to the situation. The National Health Commission held a press conference on 20 January. Zhong Nanshan academician, confirmed the existence of human-to-human transmission and confirmed that there are medical personnel were infected [66]. 23 January, early morning, Wuhan epidemic air defense headquarters issued a notice No.1, from 10 o’clock, the airport, train station, and other channels to leave Wuhan temporarily closed, Wuhan was set into the “closed city” state [67].

On this basis, combined with the search index of “COVID-19”, “pneumonia”, “pandemic” and other keywords provided by Baidu index, it can be found that the search trend of these keywords has a common feature. They all appeared a small fluctuation on 31 December 2019, but only lasted for one day. The search trend began to appear the first sign of growth on 19 January, and began to rise dramatically from 20 January, reached a peak on 25 January, the extremely high search volume and attention has continued until mid-February to level off, so this work chooses the day Wuhan closed city—23 January 2020 as the event day, defining the event window as [−3, 3], i.e., three trading days before and after the event day, a total of seven trading days; defining the estimation window as [−214, −15], a total of 200 trading days. In this work, we use the market model to estimate the expected return in the event window as follows:

$$R_{i,t}={\alpha }_0+{\alpha }_1{R}_{m,t}+\epsilon$$

(1)

where $R_{i,t}$ denotes the daily individual stock return for company i at time t considering reinvestment of cash dividends, and $R_{m,t}$ denotes the market return at time t as measured by the composite daily market return considering reinvestment of cash dividends under the weighted average of outstanding market capitalization method. By regressing the return data within the estimation window, the estimated values of the coefficients ${\alpha }_0$ and ${\alpha }_1$ can be obtained, which in turn use the coefficient estimates and the market return estimates within the event window to obtain the expected daily individual stock return for each company within the event window, and the difference with the actual return is the abnormal return ${AR}_{i,t}$:

$${AR}_{i,t}=R_{i,t}-(\widehat{{\alpha }_0}+\widehat{{\alpha }_1}{R}_{m,t})$$

(2)

Accumulating the abnormal returns during the event window yields the cumulative abnormal return for each company ${CAR}_{i,t}$ in the event of COVID-19.

$${CAR(-n,+n)}_{i,t}=\sum _{t=-n}^{+n}{AR}_{i,t}=\sum _{t=-n}^{+n}(R_{i,t}-(\widehat{{\alpha }_0}+\widehat{{\alpha }_1}{R}_{m,t}\left)\right)$$

(3)

In order to investigate whether CSR investments have a buffering effect on the shocks brought by the COVID-19, holding other factors constant, this work constructs the multiple regression model shown in Equation (4) below:

$$\begin{array}{c}{CAR}_i={\beta }_0+{\beta }_1{CSR}_i+{\beta }_2{Size}_i+{\beta }_3{Lev}_i+{\beta }_4{ROA}_i+{\beta }_5{BM}_i+{\beta }_6{Tobin}_i+{\beta }_7{Top}_i\hfill \\ +{\beta }_8{Inddirector}_i+{\beta }_9{Director}_i+{\beta }_{10}{Dual}_i+\epsilon \hfill \end{array}$$

(4)

where CSR stands for Corporate Social Responsibility Rating, this work adopts the ratings under the professional evaluation system of Hexun’s Social Responsibility Report for Listed Companies, which examines five dimensions, namely, shareholder responsibility, employee responsibility, supplier and consumer responsibility, environmental responsibility and social responsibility, with each dimension divided into 13 secondary indicators and 37 tertiary indicators, to provide a comprehensive evaluation of social responsibility. In this work, the total social responsibility score of each enterprise in 2019 is used as the CSR variable. In addition, drawing on relevant literature, a series of control variables are controlled in the model. Among them, Size represents the size of the company, Lev is the debt level of the company, ROA is the return on total assets, reflecting the profitability of the company, BM is the book-to-market ratio, Tobin is the Tobin’s Q value, reflecting the value of the enterprise, Top is the proportion of the first largest shareholder’s ownership, Inddirector represents the proportion of the sole director, Director represents the size of the directors, and Dual reflects whether the chairman of the board of directors serving as the CEO concurrently. The specific definitions of each variable are shown in

Table 1. Variable Definitions.

Variable	Definition
CAR	Cumulative abnormal return, derived using market model
CSR	Social responsibility investment, equal to the CSR scores of 2019 from the professional evaluation system of Hexun’s Social Responsibility Report for Listed Companies
Size	Firm size, equal to natural logarithm of the book value of total assets
Lev	Level of debt, equal to ratio of total debt divided by total asset
ROA	Return on assets, equal to net income divided by total assets
BM	Book-to-market ratio, equal to total assets divided by market value
Tobin	Tobin’s Q
Top	Shareholding ratio of the largest shareholder
Inddirector	Percentage of independent board members of a company
Director	Size of directors, equal to natural logarithm of the total number of directors
Dual	If the chairman of board and CEO are the same person, take 1, otherwise take 0
SOE	Nature of property rights, state-owned enterprises take 1, non-state-owned enterprises take 0
Inv_lev	Industry leverage ratio, equal to the median leverage ratio in the industry
Growth	Firm growth, equal to total asset growth rate
Fata	Fixed asset ratio, equal to net fixed assets divided by total assets
Age	Firm age, equal to 2019 minus year the firm was founded
Elev	Firms with excessive debts take 1, otherwise take 0
FC	Firms with high financing constraints take 1, otherwise take 0

below. In addition, industry fixed effects are controlled in the model.

In order to measure the level of over-indebtedness, this work draws on the treatment of Harford et al. (2009), Denis and Mckeon (2012), and Lu Zhengfei et al. (2015) [62,68,69] and utilizes the following model to predict the firm’s target debt ratio:

$$\begin{array}{c}{Levb}_{i,t}={\alpha }_0+{\alpha }_1{SOE}_{i,t-1}+{\alpha }_2{ROA}_{i,t-1}+{\alpha }_3{Inv\_lev}_{i,t-1}+{\alpha }_4{Growth}_{i,t-1}\hfill \\ +{\alpha }_5{Fata}_{i,t-1}+{\alpha }_6{Size}_{i,t-1}+{\alpha }_7{Top}_{i,t-1}\hfill \end{array}$$

(5)

The model controls seven stable and reliable factors affecting firms’ debt ratios. Among them, SOE represents the nature of property rights, Inv_lev represents the median debt ratio of the industry, Growth represents the growth rate of total assets, Fata represents the proportion of fixed assets, and the rest of the variables are consistent with the previous definitions. After predicting the firms’ target debt ratios in 2019 according to the model, the actual debt ratios in 2019 are then subtracted from the target debt ratios to obtain the level of over-indebtedness of each firm. The dummy variable Elev is set accordingly, and is taken as 1 when the over-indebtedness ratio is greater than 0, representing over-indebted companies, and 0 when it is less than 0, representing under-indebted companies.

In order to measure the degree of financing constraints, this work adopts the SA index constructed by two variables with strong exogeneity, namely, firm size and firm age, which are calculated by the following model:

$${SA}_i=-0.737{Size}_i+0.043{Size}_i^2-0.04{Age}_i$$

(6)

Age represents the age of the enterprise. The SA index is negative, and the larger absolute value represents the higher degree of financing constraints, according to which the dummy variable FC is set. When the SA index is smaller than the mean value, it is taken as 1, which represents the high financing constraints group; when the SA index is larger than the mean value, it is taken as 0, which represents the low financing constraints group.

3.2. Techniques, Sample Selection, and Data Sources

We employ multiple regression analysis to examine the relationship between the dependent variable and a set of independent variables. This method allows us to estimate the effect of each explanatory variable on the outcome variable while controlling for the influence of other covariates. All statistical analyses are conducted using Stata 17.

Our sample construction begins with A-share listed companies. A-shares refer to Renminbi-denominated ordinary shares issued by companies incorporated in mainland China and listed on the country’s two main stock exchanges: the Shanghai Stock Exchange and the Shenzhen Stock Exchange. These shares represent ownership in a broad spectrum of Chinese enterprises and are widely used in empirical research on China’s capital markets.

The corporate social responsibility (CSR) scores are obtained from Hexun.com, a prominent Chinese financial information portal and online media platform. In addition to news services, Hexun provides proprietary analytical tools, investor forums, and widely used datasets, including CSR ratings for listed companies. Other firm-level data are retrieved from the CSMAR database, a leading financial and economic database that focuses on the Chinese market. All firm-level data are selected as cross-sectional observations as of the end of 2019.

We exclude observations according to the following procedures: (1) exclude financial companies as well as ST and *ST samples (“ST” stands for “Special Treatment,” a designation assigned by Chinese stock exchanges to listed firms that are experiencing significant financial or operational difficulties, as a means of alerting investors to potential risks); (2) exclude observations with missing values in the model; and (3) exclude companies with an estimation window of less than 200 trading days. In addition, we winsorise the continuous variables at the 1% and 99% levels to deal with extreme values.

4. Empirical Results and Analysis

4.1. Descriptive Statistics

We obtain 2477 observations after the exclusion procedure, and the descriptive statistics of the main variables are demonstrated in

Table 2. Descriptive Statistics.

Variables	Observations	Mean	Standard Deviation	Min	Median	Max
CAR	2477	−0.008	0.114	−0.201	−0.035	0.465
CSR	2477	19.819	9.033	−3.760	21.340	36.500
Size	2477	22.573	1.330	19.186	22.416	26.366
Lev	2477	0.443	0.196	0.060	0.433	0.995
ROA	2477	0.029	0.091	−0.705	0.034	0.258
BM	2477	0.711	0.255	0.121	0.721	1.220
Tobin	2477	1.717	1.066	0.820	1.386	8.246
Top	2477	0.350	0.146	0.090	0.328	0.721
Inddirector	2477	0.378	0.054	0.333	0.364	0.571
Director	2477	8.479	1.658	4.000	9.000	17.000
Dual	2477	0.266	0.442	0.000	0.000	1.000

below. The mean value of CAR is −0.008, and the median is −0.035, which means that the cumulative abnormal return of the enterprises after the outbreak of COVID-19 is negative, reflecting the negative impact of COVID-19 on the market reaction, which is in line with the expectation of this work. The mean value of CSR is 19.819, the median is 21.340, and the standard deviation is 9.033, which shows that the average level of CSR of enterprises in China is slightly lower, and the difference among enterprises is obvious. The mean and standard deviation of other control variables are similar to those of previous studies, and the distribution characteristics are basically in line with normal distribution.

4.2. Basic Regression Results

4.2.1. The Buffering Effect of CSR on COVID-19 Shocks

Table 3. Regression Analysis of CSR and CAR.

	(1)	(2)
	CAR	CAR
CSR	0.0014 ***	0.0009 **
	(5.23)	(2.27)
Size		0.0135 ***
		(5.81)
Lev		−0.0406 ***
		(−2.62)
ROA		0.0231
		(0.48)
BM		−0.0347 **
		(−2.08)
Tobin		−0.0026
		(−0.77)
Top		−0.0516 ***
		(−3.27)
Inddirector		−0.0556
		(−1.27)
Director		−0.0024 *
		(−1.72)
Dual		0.0112 **
		(1.99)
_cons	−0.0368 ***	−0.2283 ***
	(−6.17)	(−5.14)
Industry FE	YES	YES
n	2477	2477
Adj-R2	0.049	0.062
Change in Adj-R2	--	0.013 ***
F-statistics	27.30 ***	7.95 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

below presents the regression results of CSR on CAR. Columns (1) and (2) show the regression results without and with control variables, respectively. It can be seen that the regression coefficients of CSR variables are significantly positive in both columns. In particular, the coefficient of the CSR variable is significantly positive at the 5% level in the regression results with control variables in column (2). In terms of economic implications, for every one-point increase in the CSR score, the cumulative abnormal return of the firm increases by about 0.1% and the market value of the firm increases by an average of ¥10,975,500, while other control variables are held constant. Therefore, we believe that the buffering impact of CSR investments is economically significant.

It is indicated that CSR has a positive impact on CAR, which means CSR investments can mitigate the negative impact of COVID-19 on stock price, so that Hl is confirmed. In addition, Firm size is significantly positive at the 1% level, indicating that the larger the firm, the more capable it is to withstand the impact of negative shocks such as COVID-19. Leverage is significantly negative at the 1% level, showing that firms with more debt are more severely affected by the negative impacts due to liquidity difficulties. The coefficient of the shareholdings ratio of the largest shareholder is significantly negative at the 1% level, which means that firms with a larger proportion of shares held by the largest shareholder will be more negatively impacted by the pandemic. The signs of other control variables are also consistent with the findings of the existing literature.

4.2.2. Comparison of Subgroups Based on Level of Over-Indebtedness

Table 4. Comparison of Different Debt Levels.

	(1) Elev = 0	(2) Elev = 1
	CAR	CAR
CSR	0.0001	0.0014 **
	(0.27)	(2.37)
Size	0.0177 ***	0.0155 ***
	(4.96)	(3.88)
Lev	−0.1047 ***	−0.0820 **
	(−3.11)	(−2.34)
ROA	0.0346	0.0014
	(0.62)	(0.02)
BM	−0.0329	−0.0357
	(−1.56)	(−1.30)
Tobin	−0.0029	−0.0015
	(−0.64)	(−0.30)
Top	−0.0409 *	−0.0670 ***
	(−1.90)	(−2.78)
Inddirector	−0.0607	−0.0528
	(−1.01)	(−0.80)
Director	−0.0010	−0.0038 *
	(−0.53)	(−1.76)
Dual	0.0101	0.0122
	(1.27)	(1.48)
_cons	−0.3074 ***	−0.2382 ***
	(−4.58)	(−3.32)
Industry FE	YES	YES
Test for coefficient	−0.0013 ***
difference (p-value)	(0.005)
n	1256	1221
Adj-R2	0.055	0.064
F-statistics	3.56 ***	6.61 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

below demonstrates a comparison of the differences in the impact of CSR on CAR between the over-indebted and under-indebted firms. Column (1) shows the results for under-indebted firms; the actual debt ratio of this group of firms is lower than the target debt ratio measured according to the model, and the coefficient of CSR is found to be insignificant. Column (2) presents the results for firms with excessive debts; the actual debt ratios of this group are higher than the target debt ratio, and the coefficient of CSR is 0.0014 and is significantly positive at the 5% level. The difference in coefficients between these two groups is significant at the 1% level. The regression results for the remaining control variables are similar to those above, and the signs are consistent with previous literature. Comparing the two columns shows that the buffering effect of CSR on CAR is more significant for over-indebted firms compared to under-indebted firms. This means that for firms with excessive debts, the buffering effect of CSR investments against the pandemic is more pronounced, so that H2 is tested. Less indebted firms retain a higher level of liquidity during COVID-19 and are able to utilize available funds to cope with the difficulties, so that CSR plays a relatively less significant role. On the other hand, over-indebted firms, which are financially inflexible, and the insurance effect brought by CSR can be more clearly exerted to mitigate the negative impact through the trust mechanism.

4.2.3. Comparison of Subgroups Based on the Degree of Financing Constraints

Table 5. Comparison of Different Financing Constraints.

	(1) FC = 0	(2) FC = 1
	CAR	CAR
CSR	0.0006	0.0015 ***
	(0.83)	(2.88)
Size	0.0137 ***	0.0151 ***
	(4.68)	(3.38)
Lev	−0.0145	−0.0583 ***
	(−0.65)	(−2.68)
ROA	0.0039	0.0275
	(0.04)	(0.59)
BM	−0.0717 ***	−0.0039
	(−2.83)	(−0.17)
Tobin	−0.0022	−0.0030
	(−0.38)	(−0.72)
Top	−0.0455 **	−0.0443 *
	(−2.08)	(−1.87)
Inddirector	−0.0845	−0.0226
	(−1.32)	(−0.38)
Director	−0.0030	−0.0018
	(−1.49)	(−0.88)
Dual	−0.0004	0.0242 ***
	(−0.06)	(2.74)
_cons	−0.1973 ***	−0.3103 ***
	(−3.63)	(−3.38)
Industry FE	YES	YES
Test for coefficient	−0.0009 *
difference (p-value)	(0.056)
n	1238	1239
Adj-R2	0.053	0.078
F-statistics	4.19 ***	7.03 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

below demonstrates the difference in the impact of CSR on CAR between the two groups with different financing constraints. Column (1) shows the group with low financing constraints, and the coefficient of CSR is 0.0006, but it is insignificant. Column (2), on the other hand, shows the group with high financing constraints, and the coefficient of CSR is 0.0015, which is positive at the 1% level. The results of the test for difference in coefficients between groups indicate that the CSR coefficients of the two groups are significantly different at the 10% level of significance. The regression results for the remaining control variables are similar to those above, and the signs are consistent with previous literature. The comparison of the results in the two columns reflects that the buffering effect of CSR investments against the pandemic is more pronounced for firms with higher financing constraints. Firms with low financing constraints are able to obtain more support to maintain normal business activities in the event of a crisis, and the role played by CSR becomes relatively less pronounced. For enterprises with high financing constraints, the social capital brought by CSR has more room to play the role of insurance, and its market response is more positive when financing channels are limited.

4.3. Cross-Sectional Analysis

To further understand the differences in the impact of CSR’s buffering effect on the adverse impact of COVID-19 on different firms, we conduct the following cross-sectional analysis:

4.3.1. Regional Trust Levels

As mentioned in the previous section, by taking their social responsibility, companies can strengthen the bonds between themselves and their stakeholders and form a foundation of social trust. In the event of an unexpected crisis, these stronger bonds can retain high-quality employees and maintain a stable supply chain and customers. At the same time, investors’ trust in the company makes them more stable in the face of unexpected events, thus cushioning the negative impact of such events on the company. Since companies build social capital through CSR investments, which in turn motivates stakeholders to remain loyal in the face of a crisis and acts as a buffer, this effect is inevitably affected by the local social trust environment. In regions with high social trust, it is easier to build trust between enterprises and stakeholders, and the positive impact of CSR can be better realized. In order to verify this relationship, this work draws on Cui Wei and He Yan’s [70] definition of regional trust levels and utilizes the China Family Panel Studies (CFPS) database of the Institute of Social Science Survey at Peking University as a data source. The survey followed all family members of the sample households, gathering data at the individual, household, and community levels to capture the shifts in China’s social, economic, demographic, educational, and health conditions. The questionnaire covered a range of questions on family and social relationships, among which the question related to trust was No.N1001: “Do you think most people can be trusted, or is it better to be more careful with people?”. In this work, using the national survey data from 2018, the responses of the survey respondents are averaged to obtain the average score of the trust level of each province. A dummy variable, Trust, is set to represent the level of regional trust. Provinces with scores higher than the median are defined as regions with a high level of trust, and are taken to be 1, while the rest of the provinces are defined as regions with a low level of trust, and are taken to be 0. The regression results obtained after grouping enterprises according to the regional trust level of their registered location are shown in

Table 6. Cross-sectional Analysis—Regional Trust Levels.

	(1) Trust = 0	(2) Trust = 1
	CAR	CAR
CSR	0.0004	0.0019 ***
	(0.82)	(2.95)
Size	0.0180 ***	0.0080 **
	(5.44)	(2.40)
Lev	−0.0772 ***	0.0044
	(−3.58)	(0.20)
ROA	0.0548	−0.0275
	(1.09)	(−0.31)
BM	−0.0255	−0.0449 *
	(−1.04)	(−1.91)
Tobin	−0.0029	−0.0030
	(−0.64)	(−0.59)
Top	−0.0779 ***	−0.0236
	(−3.78)	(−0.93)
Inddirector	−0.0594	−0.0511
	(−1.00)	(−0.76)
Director	−0.0035 *	−0.0014
	(−1.87)	(−0.65)
Dual	0.0143 *	0.0064
	(1.89)	(0.75)
_cons	−0.2917 ***	−0.1526 **
	(−4.63)	(−2.39)
Industry FE	YES	YES
Test for coefficient	−0.0015 **
difference (p-value)	(0.049)
n	1322	1153
Adj-R2	0.071	0.046
F-statistics	7.27 ***	3.81 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

.

It can be seen that the coefficient of CSR is 0.004, which is not statistically significant, in the low trust level region demonstrated in column (1), while the coefficient of CSR in the high trust level region in column (2) is 0.0019 and is significantly positive at the 1% significance level, which indicates that each one-point increase in CSR leads to a 0.19% increase in CAR by 0.19%. The difference in the coefficients is significant at 5% level. This implies that in regions with higher social trust, firms are able to acquire social capital through CSR investments, which can play a positive role in mitigating the impact of negative events on firms’ stock returns. In contrast, in areas with weak social trust, CSR investments by enterprises may be perceived as an image project, and social capital becomes more difficult to obtain, resulting in a less effective buffering effect.

4.3.2. Regional Severity of COVID-19

At the beginning of the outbreak, although confirmed cases were reported in all provinces of the country, they were mainly concentrated in Hubei and its neighboring provinces, so the severity of COVID-19 varied from region to region, and the negative impacts brought about by the outbreak also varied regionally. It is reasonable to expect that the buffering effect of CSR should be more pronounced in regions where the pandemic is more severe. To verify this idea, this work sets a dummy variable COVID to represent the severity of the pandemic, defining provinces with 600 or more confirmed cases as of 5 February 2020, the last day of the event window for CAR, as severe areas, and setting COVID as 1. Provinces with 200 or fewer confirmed cases are considered mild areas, and taking COVID as 0.

Table 7. Cross-sectional Analysis—Regional Severity of COVID-19.

	(1) COVID = 0	(2) COVID = 1
	CAR	CAR
CSR	−0.0001	0.0014 **
	(−0.13)	(2.00)
Size	0.0073	0.0170 ***
	(1.32)	(4.44)
Lev	−0.0646 *	−0.0188
	(−1.82)	(−0.80)
ROA	0.0219	−0.0243
	(0.18)	(−0.28)
BM	−0.0075	−0.0696 **
	(−0.19)	(−2.35)
Tobin	−0.0060	−0.0026
	(−0.89)	(−0.42)
Top	0.0082	−0.0577 **
	(0.21)	(−2.17)
Inddirector	0.1057	−0.1020
	(0.91)	(−1.49)
Director	−0.0003	−0.0022
	(−0.10)	(−0.95)
Dual	0.0432 **	0.0115
	(2.48)	(1.24)
_cons	−0.1809 *	−0.2849 ***
	(−1.86)	(−3.59)
Industry FE	YES	YES
Test for coefficient	−0.0015 *
difference (p-value)	(0.099)
n	452	895
Adj-R2	0.055	0.054
F-statistics	1.60	4.87 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

presents the regression results after grouping provinces by COVID-19 severity at the time of firm registration.

It can be found that there is a significant positive relationship between CSR and CAR at a significance level of 5% only in areas with severe COVID-19, whereas no such significant positive relationship exists for firms in areas with milder outbreaks, and the difference in coefficients between the groups is significant at a level of 10%. It implies that each point increase in the social responsibility score of firms in areas with severe outbreaks brings a 0.14% increase in cumulative abnormal returns, validating the previous conjecture.

4.4. Analysis of Differences in the Role of Sub-Dimensional CSR

Hexun’s professional evaluation framework for its Listed Companies Social Responsibility Report Ranking assesses firms across five key dimensions: shareholder relations, employee welfare, supplier and consumer engagement, environmental impact, and social responsibility. Among these different dimensions of social responsibility, which dimension plays a major role in buffering when a company encounters a negative disaster such as COVID-19? To answer this question, this work proposes to regress CAR on each of the five dimensions using their respective scores to determine whether each dimension has a significant buffering effect. In this case, since the supplier and consumer responsibility and environmental responsibility of each company show a score of 0 in 2019, these two dimensions are not included in the analysis.

Table 8. Regression Analysis of Sub-dimensional CSR and CAR.

	(1)	(2)	(3)
	CAR	CAR	CAR
CSR_A	0.0012 *
	(1.93)
CSR_B		−0.0012
		(−0.64)
CSR_E			0.0012 *
			(1.90)
Size	0.0136 ***	0.0153 ***	0.0147 ***
	(5.79)	(6.62)	(6.53)
Lev	−0.0391 **	−0.0471 ***	−0.0468 ***
	(−2.47)	(−3.12)	(−3.11)
ROA	0.0222	0.0776 **	0.0633 *
	(0.42)	(2.08)	(1.66)
BM	−0.0346 **	−0.0389 **	−0.0380 **
	(−2.07)	(−2.31)	(−2.26)
Tobin	−0.0024	−0.0028	−0.0029
	(−0.73)	(−0.86)	(−0.87)
Top	−0.0522 ***	−0.0463 ***	−0.0471 ***
	(−3.30)	(−2.92)	(−2.98)
Inddirector	−0.0533	−0.0569	−0.0594
	(−1.22)	(−1.30)	(−1.35)
Director	−0.0024 *	−0.0024 *	−0.0024 *
	(−1.70)	(−1.71)	(−1.74)
Dual	0.0106 *	0.0113 **	0.0118 **
	(1.89)	(2.00)	(2.10)
_cons	−0.2307 ***	−0.2453 ***	−0.2385 ***
	(−5.20)	(−5.53)	(−5.41)
Industry FE	YES	YES	YES
n	2477	2477	2477
Adj-R2	0.061	0.059	0.061
F-statistics	8.22 ***	7.06 ***	7.31 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

below presents the results of the regression of the scores of the shareholder responsibility CSR_A, employee responsibility CSR_B, and social responsibility CSR_E dimensions on the CAR.

The regression analysis reveals statistically significant coefficients of 0.0012 for both shareholder responsibility and social responsibility, and both are significant at the 10% level, while the regression coefficient of employee responsibility is insignificant. The regression coefficients and signs of the remaining control variables and their significance levels are also largely consistent with similar literature. The regression results show that the buffering effect of CSR on negative pandemic shocks is mainly manifested through the dimensions of shareholder responsibility and social responsibility. Firms with higher shareholder responsibility scores imply more positive performance in profitability, debt servicing, returns, and innovation, while firms with higher social responsibility scores represent a greater value of their contribution to society, and the resulting financial stability and reputation solidify the link between the firm and its stakeholders, win their trust, and are thus able to maintain their corporate value and stability in the face of negative shocks such as the COVID-19.

4.5. Robustness Tests

To strengthen the validity of our core findings, we conduct the following robustness checks:

4.5.1. Change CSR Definition

In the previous part, the score of 2019 was used as the CSR measure. In order to test the robustness of the conclusion, this definition is replaced with the average of the scores of 2017–2019 in the professional evaluation system of Hexun to portray CSR, and the obtained results are briefly presented in column (1) of

Table 9. Robustness test—Changing CSR Definition and Event Window.

	(1)	(2)	(3)
	CAR	CAR	CAR
CSR	0.0008 *	0.0010 ***	0.0009 **
	(1.92)	(3.10)	(2.06)
_cons	−0.2274 ***	−0.2680 ***	−0.2135 ***
	(−5.08)	(−7.25)	(−4.17)
Controls	YES	YES	YES
Industry FE	YES	YES	YES
n	2477	2477	2477
Adj-R2	0.061	0.069	0.069
F-statistics	7.95 ***	7.95 ***	7.95 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

. After replacing the CSR variable with the mean, it remains positively correlated with CAR and is significant at the 10% level. Firms with higher CSR means have better market reactions in the event of COVID-19, which supports the findings of H1. Meanwhile, in the unreported results of comparisons between different levels of indebtedness and different degrees of financing constraints, the CSR variable remains significantly positive only in the excessive debts and high financing constraints groups, and the conclusions of H2 and H3 still hold significantly, with good robustness.

4.5.2. Change Event Window

Columns (2) and (3) in Table 9 present the main regression results after changing the event window to [−2, 2] and [−5, 5], respectively. The coefficient signs and significance presented in the table still support the conclusions of H1 in the previous section, with CSR being significantly positive at the 1% and 5% levels, respectively. In the unreported results comparing different levels of indebtedness and different degrees of financing constraints, the CSR variable remains significantly positive only in the excessive debts and high financing constraints groups, and the conclusions of H2 and H3 still hold significantly. This suggests that changes in the event window do not affect the main conclusions in the work. The conclusion that CSR can act as a buffer against shocks from negative crises and that this effect is more pronounced for over-indebted and highly finance-constrained firms is robust.

4.5.3. Change Grouping Method

In order to further demonstrate the robustness of the conclusions of H2 and H3, this work adjusts the grouping methods and re-tests them while keeping the variable definitions unchanged. While in the previous part the observations were divided into two groups based on the mean values, in the robustness test, three quartiles are utilized, defining the top third of the over-indebtedness ratio as the over-indebtedness group and the bottom third of the companies as the under-indebtedness group, and the top one-third of the absolute value of the SA index as the high financing constraints group and the bottom one-third as the low financing constraints group. The regression results obtained are briefly presented in

Table 10. Robustness Test—Changing Grouping Methods.

	(1) Elev = 0	(2) Elev = 1	(3) FC = 0	(4) FC = 1
	CAR	CAR	CAR	CAR
CSR	0.0002	0.0020 ***	−0.0004	0.0017 ***
	(0.35)	(2.82)	(−0.57)	(2.74)
_cons	−0.3664 ***	−0.2398 **	−0.1673 ***	−0.2706 **
	(−4.30)	(−2.50)	(−2.73)	(−2.16)
Controls	YES	YES	YES	YES
Industry FE	YES	YES	YES	YES
Test for coefficient	−0.0018 ***		−0.0021 *
difference (p-value)	(0.003)		(0.054)
n	826	825	825	826
Adj-R2	0.056	0.066	0.044	0.079
F-statistics	3.56 ***	6.01 ***	4.19 ***	7.03 ***

Note: Values reported in parentheses are t-statistics, with standard errors adjusted for heteroskedasticity; Significance levels are indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.01, respectively.

. Consistent with the findings obtained in the previous section, the CSR remains significant and positive only in the over-indebtedness and high financing constraints groups at a significance level of 1%, while it is not significant in the remaining two groups. This indicates that CSR has a more significant buffering effect during COVID-19 for firms that have excessive debts and face higher financing constraints, and the robustness of H2 and H3 is verified.

5. Discussion

5.1. Key Findings

This study investigates how and when CSR buffers firms against adverse shocks. The findings contribute to the research on signaling theory and CSR by revealing three key insights.

First, the results demonstrate that firms with higher CSR investments experienced significantly greater cumulative abnormal returns during the pandemic. CSR is an effective way to build trust by conveying a credible and visible signal.

Second, the study finds that firms’ financial situation and regions’ background are key moderators that enhance CSR’s buffering effect. Specifically, when firms have higher debt burdens, greater financing constraints, and regions where they operate have stronger social trust and more severe COVID-19 impact, the buffering effect is magnified. These results align with prior research [57,63], which emphasizes the roles of internal strategic imperatives and external environmental pressures in strengthening the effectiveness of CSR’s resilience mechanism.

Third, the study highlights the complex, multifaceted mechanisms of the buffering effects of CSR investments, which solved the paradoxes of prior research [26,27,28,29]. Existing researches mainly focus on whether CSR has a buffering effect. In the process of studying the CSR mechanism, we took into account how CSR serves as a resilience mechanism in adverse environments and when it proves most effective.

5.2. Theoretical Contributions

Our study aims to make the following theoretical contributions.

First, by integrating signaling theory with crisis management literature, our research re-frames CSR not merely as a reactive or symbolic tool for legitimacy, but as a proactive strategic asset that buffers firms against adverse impacts during crises. This perspective transcends the conventional view of CSR as a passive reputation-enhancing mechanism, instead positioning it as a forward-looking strategic signal that communicates organizational stability, stakeholder orientation, and long-term value commitment. In doing so, we contribute to the evolving nature of CSR as a dynamic resource with instrumental value in times of systemic turbulence.

Second, we advance understanding of the mechanism through which CSR operates in the context of pandemic-induced crises. Specifically, we uncover the transmission channel by which CSR alleviates market instability: by enhancing information transparency and stakeholder trust, CSR reduces information asymmetry and boosts investor confidence. This mechanism highlights CSR’s communicative and informative role, offering a nuanced answer to the critical question of how CSR functions as a resilience mechanism under adverse conditions.

Third, our findings emphasize the contingent nature of CSR’s efficacy, showing that CSR initiatives are most effective when they align with both internal strategic imperatives and external environmental pressures. This dual contingency framework underscores the importance of contextual fit, illustrating that CSR’s signaling power is activated most strongly when firms are simultaneously navigating internal alignment challenges and responding to exogenous shocks. In doing so, our study extends signaling theory by demonstrating its relevance in high-uncertainty, non-financial disclosure contexts, where signals related to ethics, responsibility, and stakeholder engagement become critical for organizational resilience. This insight helps answer the research question of when CSR proves most effective, offering a context-sensitive extension of CSR’s strategic function during systemic crises.

5.3. Practical Implications

The findings of our study also provide practical implications.

For firms, the findings underscore the strategic importance of CSR investments not only as ethical or reputational commitments but as proactive resilience mechanisms that can be mobilized during times of systemic crisis. By leveraging CSR as a credible and visible signal, firms can enhance stakeholder trust, reduce uncertainty, and mitigate adverse market reactions. This signaling effect is particularly valuable when traditional performance indicators are obscured or volatile, such as during global health emergencies or economic recessions. Importantly, in today’s increasingly ideologically polarized and geopolitically volatile world, our results suggest that CSR should no longer be viewed as a discretionary or peripheral activity. Instead, it should be embedded into a firm’s core strategic formulation and implementation, serving as a source of core competence and a foundation for sustainable competitive advantage. Moreover, the effectiveness of CSR in crisis contexts is contingent on both internal alignment and external responsiveness. Thus, managers should design CSR strategies that reflect organizational values while remaining sensitive to evolving stakeholder expectations and environmental challenges. This strategic alignment enhances the authenticity and credibility of CSR signals, increasing their support from key stakeholders such as investors, customers, employees, and regulators.

For policymakers and regulatory agencies, our study highlights the critical role of institutional support in amplifying the positive externalities of CSR. In particular, fostering regulatory and normative environments that encourage CSR engagement can enhance corporate transparency and promote broader system-level resilience. Policy instruments such as tax incentives, preferential access to public resources, public recognition programs, and the incorporation of CSR criteria into corporate governance standards can incentivize firms to adopt socially responsible practices in a sustainable and substantive manner.

5.4. Chinese-Context Insights

Critically, this Chinese experience offers three transferable insights for global scholars and policymakers: First, in institutionally fragmented markets (e.g., emerging economies), CSR signaling requires state-business co-governance—a core Chinese feature where local governments amplify CSR visibility through political mobilization while firms deliver hyper-localized authenticity. This co-governance model differs from Western market-centric approaches. Second, Chinese crisis-responsive policy engineering demonstrates how infrastructure-level institutionalization can accelerate positive externalities. Third, the “dual contingency” framework (internal financial constraints and external trust/pressure) provides a universal diagnostic tool: firms in debt crises or high-trust regions globally should prioritize CSR as a strategic signal, not just an ethical gesture. Future research should test these mechanisms in varied institutional settings.

5.5. Limitations and Future Research

While our findings offer important insights, several limitations open avenues for future research.

First, the buffering effect of CSR appears context-dependent—shaped by institutional factors such as social trust and the intensity of the crisis—suggesting that its efficacy is not universal. Our geographic focus on China may limit the generalizability of results to other institutional and cultural contexts. Therefore, cross-national comparative studies are needed to disentangle the institutional and cultural contingencies that moderate CSR’s effectiveness.

Second, although the event study design effectively captures short-term market reactions, it does not address the long-term impact of CSR during protracted or recurring crises.

Future research should adopt longitudinal designs to explore how CSR’s resilience function evolves across different crisis phases. Furthermore, future studies could explore the interplay between CSR and other intangible assets—such as innovation capacity or digital capabilities—to identify potential synergistic effects in enhancing crisis resilience.

6. Conclusions

This study employs the event study methodology with A-share listed company data to examine CSR’s buffering effect on market reactions during COVID-19, investigating how CSR serves as a resilience mechanism in adverse environments and when it proves most effective. The study finds that firms with higher CSR scores have a more positive market reaction during the pandemic, suggesting that CSR can play a buffering role against the negative shocks of an unexpected crisis like COVID-19, and this buffering role is more pronounced for firms with excessive debts and firms facing high financing constraints. Cross-sectional analysis across regions indicates that the buffering effect of CSR is more pronounced in regions with high levels of trust and severe outbreaks of COVID-19. These findings hold after robustness tests.

By doing so, our study enriched the literature around CSR’s role in crisis contexts and enhanced our understanding of its dual function as both a communicative signal and a strategic resilience mechanism. Theoretically, this study reframes CSR as a proactive strategic asset transcending symbolic legitimacy by integrating signaling theory with crisis management, revealing dual mechanisms: as a communicative signal reducing market instability through enhanced transparency/stakeholder trust, and as a contingency-activated resource maximized when internal strategic imperatives align with external pressures, extending signaling theory to high-uncertainty non-financial contexts. Practically, firms must embed authentic CSR into core strategy as a competence-building resilience lever, while policymakers should deploy institutional engineering to amplify positive externalities. Limitations include China’s context requiring cross-national validation, short-term focus needing longitudinal crisis-phase analysis, and unexplored digital synergies.